# https://github.com/AustinBlues/RSS-Speed-Reader/blob/master/lib/rss_speed_reader.rb
# https://github.com/tenderlove/nokogiri/pull/524
require 'mini_exiftool'

module Daengine::TeamsiteMetadataParser

  @@translation = {
      'TeamSite/Metadata/rttTitle' => 'title',
      "TeamSite/Metadata/enterprise_last_updated_date"=> 'changed_at',
      "TeamSite/Metadata/enterprise_audience_id"=> 'audiences',
      "TeamSite/Metadata/enterprise_sami_desc"=> 'sami_code',
      "TeamSite/Metadata/enterprise_last_publication_date"=> 'published_at',
      "TeamSite/Metadata/enterprise_unpublish_date"=> 'unpublished_at',
      "TeamSite/Metadata/enterprise_expiration_date"=> 'expires_at',
      "TeamSite/Metadata/enterprise_guid" => 'guid',
      "TeamSite/Metadata/shortSynopsis" => 'summary',
      "TeamSite/Metadata/business_owner" => 'business_owner',
      "TeamSite/Metadata/enterprise_product_id" => 'product_ids',
      "TeamSite/Metadata/enterprise_content_organization_id" => 'content_organization_ids',
      "TeamSite/Metadata/enterprise_program_id" => 'program_ids',
      "TeamSite/Metadata/omnitureSiteSection_codes" => 'omniture_codes'
  }
  @@path_tuples = {
      "path" => 'path',
      "TeamSite/Metadata/enterprise_last_content_update_date" => 'doc_changed_at',
      "TeamSite/Metadata/enterprise_content_type_id" => 'content_type'
  }
  @@validations = {
      "TeamSite/Metadata/display_on_website" => lambda { |val| /Y|1|true/ =~ val },
      "TeamSite/Metadata/enterprise_expiration_date" => lambda {|val| !val.blank? },
      # "TeamSite/Metadata/enterprise_unpublish_date" => lambda {|val| val.blank? },
      #"path" => lambda {|val| !(/\/manifest\// =~ val) }
  }

  @@logger = nil

  def self.logger=(some_logger)
    @@logger = some_logger
    self
  end

  def self.log(args)
    @@logger.error(args) unless @@logger.blank?
  end

  def self.parse_tuple_file(file, last_read = nil)
    time do
      asset = nil
      assets = {}
      docpath = {}
      valid = true
      while (line = file.gets)
        case line
          when /<\/?data-tuple>/
            if (asset.blank?)
              asset = DigitalAsset.new
            elsif (valid)
              assets[asset.guid] ||= asset.attributes # first tuple metadata wins
              assets[asset.guid]['documents_attributes'] ||= []
              assets[asset.guid]['documents_attributes'] << docpath
              # assets[asset.guid]['_id'] = asset.guid
              asset = nil; docpath = {}; valid = true;
            else
              asset = nil; docpath = {}; valid = true;
            end
          when /<tuple-field name="([^"]+)">([^<]+)<\/tuple-field>/
            if (valid)
              if @@validations[$1]
                valid = @@validations[$1].call($2)
              end
              if @@path_tuples[$1]
                docpath[@@path_tuples[$1]] = $2 # if this is one of our keys, 'send' to the method
              elsif (@@translation[$1])
                val = asset.send("#{@@translation[$1]}").respond_to?(:[]) ? $2.split(',') : $2
                method_name = "#{@@translation[$1]}="             
                val = val.gsub(/(\w{3} \d{2} \d{4} \d{2}:\d{2}:\d{2}):\d{3}(AM|PM)/,'\1 \2') if val.respond_to?(:gsub) # deal with strange teamsite date formats
                asset.send(method_name, val)
              end
            end
        end
      end
      # loop thru each doc in the collection, either replace or delete it
      error_files = []
      update_count = 0; delete_count = 0; added_count = 0;
      assets.keys.each do |key|
        da = nil
        begin
          if (!assets[key]['unpublished_at'].nil?)
            DigitalAsset.where(guid: key).try :delete_all
            delete_count += 1
          elsif (assets[key]['expires_at'].nil? || (assets[key]['expires_at'] < 1.minute.from_now))
            DigitalAsset.where(guid: key).try :delete_all
            delete_count += 1
          else
            da = DigitalAsset.find_or_initialize_by(guid: key)
            asset_docs = trim_package(assets[key]['documents_attributes'], last_read, da)
            if (!asset_docs.empty?)
              assets[key]['documents_attributes'] = asset_docs
              assets[key]['orderable'] = da.orderable unless da.new?

              creating = da.new?
              da.documents = []
              begin
                da.update_attributes!(assets[key])
              rescue Exception => e
                error_files << "#{e} ----  #{da.try(:guid)}, #{da.try(:errors).try(:full_messages)}"
              end
              creating ? added_count += 1 : update_count += 1
            else
              DigitalAsset.where(guid: key).try :delete_all
              delete_count += 1
            end
          end
        rescue Exception => e
         Daengine.log(e.message, "error") 
         Daengine.log(e.backtrace.join('\n'), "error")
        end
      end
      log_txt = "TeamsiteMetadataParser: Failed to save/update following DigitalAssets in database:\n"
      error_files.each do |asset_file|
        log_txt << "> #{asset_file}\n"
      end
      Daengine.log(log_txt, "warn") unless error_files.empty?
      Daengine.log("checking purge condition, stale count is #{DigitalAsset.stale.count}, bulk_processed is #{DigitalAsset.bulk_processed?}", 'warn')
      DigitalAsset.purge! # if the purge criteria is met, purge anything not updated
      Daengine.log("TeamsiteMetadataParser: #{added_count} records added, #{update_count} updated, #{delete_count} removed", "info")
    end
  end

  def self.trim_package(asset_docs, last_read = nil, existing_da = nil)
    docs = []
    path = Daengine.config[:digital_assets_file_directory]
    if Dir.exist?(path)  # dont throw away assets if we can't locate the dir
      asset_docs.each do |doc|
        #exclude manifest_file
        unless doc['path'].match('\/manifest\/')
          file = File.join(path, doc['path'])
          if(!File.exist?(file)) # try with just the filename
            #Daengine.log "Cant find file #{file}, trying with just  the path and filename", "info"
            file = File.join(path, doc['path'].split('/').last)
          end
          #exclude digital_asset_files that are not in *Teamsite Staging*
          if File.exist?(file)
            docs << doc
            if(File.mtime(file) > last_read)
              begin
                exifdata = ::MiniExiftool.new(file) # spaces in filename
                pages = exifdata.pagecount
                pages = exifdata.pages if pages.blank?
                pages = exifdata.slides if pages.blank?
                if(pages.blank?)
                  Daengine.log "Unable to get pages metadata for #{file}, exifdata.pagecount = #{exifdata.pagecount}, exifdata.pages = #{exifdata.pages}", "info"
                end
                doc['pages'] = pages
                doc['size'] = exifdata.filesize
                doc['mime_type'] = exifdata.mimetype
                #doc['keywords'] = exifdata.keywords.gsub(';', ',').gsub(':', ',').split(",") unless exifdata.keywords.nil?
                if exifdata.keywords.is_a? Enumerable
                  doc['keywords'] = exifdata.keywords
                else
                  doc['keywords'] = exifdata.keywords.gsub(';', ',').gsub(':', ',').split(",") unless exifdata.keywords.nil?
                end
                if exifdata.description.is_a? Enumerable
                  doc['subject'] = exifdata.description.join(" ") unless exifdata.description.nil?
                else
                  doc['subject'] = exifdata.description.gsub(':', '') unless exifdata.description.nil?
                end
                doc['author'] = exifdata.author
                Daengine.log "Exif data for #{file} was pages:#{doc['pages']}, size:#{doc['size']}", "info"
              rescue Exception => e
                Daengine.log "Error reading metadata from #{file} #{e.message}", "error"
              end
            elsif (existing_da && !existing_da.new?) # copy over the existing asset metadata
              existing_doc = existing_da.documents.where(path: doc['path'])[0]
              if (existing_doc)
                doc['pages'] = existing_doc.pages
                doc['size'] = existing_doc.size
                doc['mime_type'] = existing_doc.mime_type
                doc['keywords'] = existing_doc.keywords
                doc['subject'] = existing_doc.subject
                doc['author'] = existing_doc.author
              else
                Daengine.log "Unable to find document with path #{file} on existing asset #{existing_da.guid} title: #{existing_da.title}", "info"
              end
              Daengine.log "Saving existing pages #{existing_doc.pages} and size #{existing_doc.size} on #{file}", "info"
            end
          else
            # the file was missing on disk, show a warning!
            Daengine.log("TeamsiteMetadataParser: Unable to locate file #{file} on disk!  Removing from metadata", "warn")
          end
        end
      end
    else
      Daengine.log("******************", "error")
      Daengine.log("ERROR!!!  Unable to locate physical teamsite files at #{path}", "error")
      Daengine.log("******************", "error")
    end
    docs
  end

  def self.time
    start = Time.now
    yield
    self.log "elapsed time was #{Time.now - start}"
  end

end